IBM insists AI didn't kill software deals, just delayed them

IBM spent much of its Q2 earnings call playing defense, trying to convince investors that the software slump behind one of its worst stock drops in years was a timing problem, not a trend. CEO Arvind Krishna's pitch was simple: enterprise customers didn't stop buying software, they just blew their budgets on servers, storage, and memory to feed AI projects first. He told analysts that roughly a third of the large capex deals that stalled in Q2 have already closed in the first three weeks of the new quarter, calling it "a good indication, not yet full evidence" that this was deferral rather than destruction.

Wall Street wasn't fully buying the reassurance. Evercore's Amit Daryanani pressed Krishna on the question everyone's been asking since the preliminary numbers landed: are these dollars coming back, or gone for good. Krishna held his line, insisting the spending got redirected toward AI infrastructure rather than cut altogether.

Having spent half the call defending the past, IBM used the second half to pitch the future. Krishna argued that as enterprises pour money into AI models and infrastructure, pressure will mount to actually extract value from it, and that value will migrate toward orchestration and data layers, areas IBM clearly wants to own.

That framing sets up Project Lightwell, IBM's freshly launched service for cleaning up aging open source code. Krishna credited Anthropic's Mythos release in April with dramatically speeding up AI-driven discovery of vulnerabilities buried in legacy software that community maintainers stopped patching years ago. IBM is charging $1 million a year for Lightwell, and Krishna rattled off an impressive early client list: Bank of America, Citi, Goldman Sachs, JPMorgan Chase, Mastercard, Morgan Stanley, Visa, and Wells Fargo all reportedly signed on.

The pitch amounts to IBM selling banks a fix for a problem that AI itself is uncovering faster than ever, a tidy multibillion-dollar loop if it works. Krishna is betting that this new opportunity will outweigh whatever software revenue got pushed into later quarters, even if the jury on that deferred spending is still very much out.